Computing service providers may operate server-based (e.g., cloud-based) systems that host electronic content items, such as video games, that are accessed by large numbers of players or other users. In some examples, content item instances may be executed using a number of virtual machines operated by a computing service provider. Customers of a service provider, such as game developers, may often pay a fee for having their video games or other content hosted on the service provider's systems. This fee may often be based various factors including, for example, a number of content item instances that are allocated to a customer for executing the customer's content over time. Thus, in order to help reduce costs for a customer, it may sometimes be advantageous to attempt to allocate to the customer, at any given time, as few content item instances as necessary in order to satisfy the users of the customer's hosted content. However, attempting to reduce or minimize a customer's number of allocated content item instances while also satisfying users may often prove to be a difficult and frustrating task that involves many trade-offs. For example, customers may often attempt to reduce costs by requesting fewer allocated content item instances, but this may result in situations in which users logging-on to the system are forced to experience long wait times in order for an instance of the content item to become available. This may reduce user satisfaction, lower the user's enjoyment and appreciation of the content, and reduce demand. By contrast, in some examples, customers may attempt to improve user satisfaction by requesting greater quantities of allocated instances, but this may result in increased costs and reduced efficiency, for example when allocated instances are not being used to an appropriate capacity.